Created attachment 30065 [details] screenshot of the desktop Can't tell if it's a regression because of other bugs, so this one might have been here for a long time and it's hard to trigger it. Here's what I have so far : - the corruption almost always appear in text - I've seen slight corruption in firefox's favicons once or twice - forcing a redisplay of the window makes the corruption go away xf86-video-intel: 94fc93d4e2b88565dca17f72903d8991213c9ee8 libdrm: 2.4.14 xorg-server: 1.6.4.901 kernel: 2.6.31-rc9-00143-g8542a0b (drm-intel-next a few days before .32) Thanks
Oh and I forgot, sometimes the corruption will be a pattern, sort of like a colored staircase... I'll try to take a snapshot next time it happens. Thanks
Created attachment 30338 [details] Possibly related corruption.
Created attachment 30339 [details] Report of similar artefacts on an i945
Confirming, I've seen both types of corruptions. Thanks
21:59 < Maxi> ickle, remi`: I haven't seen any font corruption since activating DebugFlushCaches.
Hopefully this commit flushes the caches correctly: commit 9a2c18fb92659d57741bfdcacbe4f69aab361532 Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Sun Nov 29 21:07:45 2009 +0000 batch: Emit a 'pipelined' flush when using a dirty source. Ensure that the render caches and texture caches are appropriately flushed when switching a pixmap from a target to a source. This should fix bug 24315, [855GM] Rendering corruption in text (usually) https://bugs.freedesktop.org/show_bug.cgi?id=24315
Ok, found another silly little bug that could be related. Apparently we were not resubmitting batches when interrupted by a signal: libdrm: commit b73612e4fd69565aa2c5c2e9677f3e0af1945f7d Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Wed Dec 2 12:58:00 2009 +0000 intel: Repeat execbuffer after EINTR EAGAIN cannot be raised by the current code, but the system call maybe interrupted and so return EINTR. So... How close are we to finally resolving this? :)
I think this bug should be fixed now. (I have a small glimmer of doubt about the validity of not having to flush with the 2D use of the render cache...)
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.